The present invention relates to an IP router, a communication system and a band setting method used therein and its program, and in particular, to a method for controlling traffic changed by a selection of a mirror server in the Internet.
In the Internet, in order to prevent services, to be provided by a server for providing contents or programs, from being halted even when the server is down, there may be a case that the contents or programs to be provided by the server are always synchronized, and backup servers for substituting the server, when failed, are provided, conventionally.
When a content is provided by a server, or a program is downloaded from a server, it causes a load to the server if the provision or the download is performed by only one server, and also it is inefficient. Considering this fact, there may be a case that a plurality of servers, capable of providing or accommodating download of the same contents or programs, are provided.
In this case, the originating server is called an original server, and servers, synchronizing the contents or programs which can be provided by or downloaded from the original server, are called mirror servers. In the case of a plurality of mirror servers being provided as described above, an optimum server selecting device, for selecting the optimum server among a plurality of mirror servers appropriate for a request (content, program or the like) from a client, may be equipped (see, for example, the Japanese Patent Application Laid-open No. 10-312365, the Japanese Patent Application Laid-open No. 2001-331398).
In the conventional Internet aforementioned, in the case that the optimum server selecting device is equipped for selecting from a plurality of mirror servers, the optimum server selecting device is a device independent of other devices, so that there is no means that the optimum server selecting device recognizes changes in the network topology.
In order to select the optimum server even when there is any change in the network topology, a health check, for obtaining information about a running/stopping state, RTT (Round-Trip Time), throughput or the like for each server which can be an index for selecting the optimum server, must be regularly performed to thereby make the check result reflected in the selection criteria.
However, if the health check is regularly performed, there arises a problem that traffic caused by the health check presses the network. Further, even though the optimum server selecting device is capable of recognizing changes in the network topology and, with an alteration of the selection criteria of the optimum server corresponding to the changes, the traffic is changed and accordingly the required band for each route is changed, routing processing can not immediately accommodate the change.